E-cadherin expression in the epithelial components of mammary phyllodes tumors

SummaryPhyllodes tumors are rare but clinically important fibroepithelial tumors of the breast. Both epithelial and stromal components actively interact with each other to participate in phyllodes tumor development. Accumulated evidence suggests that the Wnt signaling pathway is important in this stromal-epithelial interaction. Given that Wnt signaling also affects E-cadherin–dependent cellular adhesion and alteration of E-cadherin is common in epithelial cancers, it is possible that alteration of E-cadherin occurs also in the epithelial components of phyllodes tumor. We assessed epithelial E-cadherin expression in 155 phyllodes tumor cases, including 92 benign (59%), 42 borderline (27%), and 21 malignant phyllodes tumor (14%), by immunohistochemistry. Its expression was correlated with clinicopathologic features and phyllodes tumor recurrence. Significant correlations of both membranous and cytoplasmic E-cadherin expression were found with stromal cellularity (P = .009 and .013, respectively), overgrowth (P = .005 and .009, respectively), and mitotic counts (P = .023 and .029, respectively) but not tumor grade, margin, and nuclear atypia. Interestingly, a significantly higher level of cytoplasmic epithelial E-cadherin expression was found in those tumors with recurrence (score, 278.79 ± 40.91 versus 250.00 ± 63.46) and shorter specific disease-free survival (172.24 ± 12.63 versus 207.24 ± 19.71 months). Further multivariate analysis showed epithelial E-cadherin expression as an independent prognostic factor for phyllodes tumor–specific survival (P < .001 for cytoplasmic staining and .001 for membranous staining). In conclusion, we have demonstrated an association of epithelial E-cadherin expression with stromal histologic features and disease recurrence in phyllodes tumor. These findings provide further evidence of the importance of stromal-epithelial interactions in phyllodes tumors and highlight the potential value of epithelial components in prognostication.